Overexpression of Hyaluronan Synthase 2 Alters Hyaluronan Distribution and Function in Proximal Tubular Epithelial Cells
Wisam Selbi*,
Anthony J. Day,
Marilyn S. Rugg,
Csaba Fülöp,
Carol A. de la Motte,
Timothy Bowen*,
Vincent C. Hascall and
Aled O. Phillips*
* Institute of Nephrology, Cardiff University School of Medicine, Cardiff, Wales; MRC Immunochemistry Unit, Department of Biochemistry, University of Oxford, Oxford, United Kingdom; Section of Connective Tissue Biology, Department of Biomedical Engineering; and Department of Immunology, The Cleveland Clinic Foundation, Cleveland, Ohio
Address correspondence to: Prof. Aled O. Phillips, Institute of Nephrology, University of Cardiff School of Medicine, Heath Park, Cardiff CF14 4XN, UK. Phone: +44-2920-748411; Fax: +44-2920-748470; phillipsao{at}cf.ac.uk
Received for publication August 24, 2005.
Accepted for publication March 27, 2006.
The functional consequences of increased renal cortical hyaluronanthat is associated with both acute injury and progressive scarringare unclear. The aim of this study was to characterize hyaluronansynthase-2 (HAS2)-driven HA synthesis and determine its effecton renal proximal tubular epithelial cell (PTC) function, becausethis is known to be the inducible form of HA synthase in thiscell type. Overexpression of HAS2 mRNA increased HA generation,which in the supernatant predominantly was HA of large molecularweight, whereas there was an increase in low molecular weightHA in cell-associated fractions. This was associated with increasedexpression of hyaluronidases, inhibition of HA cable formationconcurrent with reduction in HA-dependent monocyte binding,and increased pericellular HA matrix. Overexpression of HAS2led to enhanced cell migration. HA can be modified by the covalentattachment of heavy chains that are derived from the serum proteininter–-inhibitor (II), a process that is known to be catalyzedby TNF-–stimulated gene 6 (TSG-6; an inflammation-associatedprotein). Enhanced migration was abrogated by blocking antibodiesto either II or TSG-6. Addition of recombinant full-length TSG-6(TSG-6Q) or TSG-6Q_Y94F, a mutant variant with impaired HA binding,increased cell migration. Both of these proteins were able tomediate the covalent transfer of heavy chains, from II and pre–-inhibitor,onto HA. Addition of the isolated TSG-6–Link module (Link_TSG-6),which binds HA but is unable to form covalent complexes withII/pre–-inhibitor, had no effect on migration, suggestingthat TSG-6–mediated formation of heavy chain–HAcomplexes is critical in the formation of a pericellular HAmatrix.
Hyaluronan (HA) is a ubiquitous connective tissue glycosaminoglycanthat in vivo is present as a high molecular mass component ofmost extracellular matrices. Although HA is not a major constituentof the normal renal corticointerstitium (1), it is expressedaround renal proximal tubular epithelial cells (PTC) after bothacute and chronic renal injury that is caused by numerous diseases(2–5). Furthermore, increased deposition of interstitialHA correlates with both proteinuria and renal function in progressiverenal disease (6). The functional significance of alterationsin HA synthesis in the renal tubulointerstitium, however, isunclear.
Our work has focused on the role of the epithelial cells ofthe proximal tubule in the initiation of fibrosis. It is wellestablished that one mechanism by which PTC may contribute todisease progression is through transdifferentiation, a processthat involves loss of cell–cell contact, disruption ofthe tubular basement membrane, acquisition of a fibroblast phenotype,and migration of cells into the interstitium (7). We have demonstratedthat binding of HA to its principle receptor, CD44, activatesthe mitogen-activated protein kinase (MAPK) pathway and enhancesPTC migration, a process that is implicated in epithelial cell–fibroblasttransdifferentiation and progressive renal fibrosis. There isoverwhelming evidence implicating TGF-1 in progressive renalfibrosis. It is particularly important in regulation of cell–celladhesion (8) and regulation of epithelial and fibroblast cellphenotype (9,10). Synergistic effects with other cytokines (11)and also alterations extracellular matrix (ECM) (12) are knownto facilitate these changes. We previously demonstrated thatTGF-1 itself has antimigratory effects on PTC (13). Interactionof HA with CD44 and cross-talk between the CD44 and TGF- receptorincrease trafficking of TGF- receptors to lipid raft–associatedpools (14), facilitates increased receptor turnover (15), andresults in the inhibition of the antimigratory effects of TGF-1(16).
Three separate genes for human HA synthase (HAS) have been clonedand characterized: HAS1, HAS2, and HAS3 (17). We previouslyexamined the regulation of HA synthesis by renal PTC in vitro(18) and demonstrated that increased HA synthesis was associatedwith transcriptional activation of HAS2 by stimuli that areimplicated in the pathogenesis of renal injury, such as elevatedglucose concentrations and the proinflammatory cytokine IL-1(18). HAS3 mRNA was constitutively expressed by PTC but wasnot altered by stimuli that increase HA synthesis. HAS1 mRNAexpression, however, was not detected in PTC. These data suggestthat isoform-specific HA generation may have a specific effecton PTC function.
Several cell types surround themselves in vitro with HA in anorganized pericellular matrix, or "coat," that is associatedwith cell migration (19,20). Recently, in addition to theseHA coats, we demonstrated that PTC form pericellular HA cable-likestructures that bind mononuclear leukocytes via their cell surfaceCD44 receptors (21) and that binding of monocytes to these structuresattenuates monocyte-dependent generation of TGF-1 by PTC (22).These data therefore suggest that HA may be either disease promotingor disease limiting. This may depend on the context in whichHA is generated, which may in turn dictate the structural architectureinto which the HA is assembled and its functional effects. Wehypothesize that regulation of HA assembly into pericellularstructures, therefore, represents another important mechanismthat regulates PTC function.
The wide range of biologic actions of HA is postulated to derivein part from its interaction with a wide number of HA-bindingproteins, termed hyaladherins, which can be intracellular, secreted,or on the cell surface. We postulate that the hyaladherins ofthe inter–-inhibitor (II) family of serum proteoglycansand the protein product of TNF-–stimulated gene 6 (TSG-6)are critical to the organization of pericellular HA-based structures.The II family includes four plasma proteins: Free bikunin, II,pre–-inhibitor (PI), and inter–-like inhibitor (ILI)(23). Each of the last three proteins exists as a distinct assemblyof one bikunin chain with one or more unique heavy chains (HC),designated HC1, HC2, and HC3. Although the components of theII family are synthesized predominantly in hepatocytes, we previouslydemonstrated that PTC generate the PI variant of the II family(24). Much of the work on the associations of II-related proteinswith ECM has been done in studies of the cumulus cell-oocytecomplexes (25). In this system, it is clear that HA is the majorstructural component that determines the viscoelastic propertiesof the expanding matrix of the cumulus-oophorus. It is wellestablished that members of the II family are pivotal for theformation of the cell-oocyte complexes matrix, where covalentlinkage of HC of II (HC1 and HC2) and PI (HC3) to HA has beendemonstrated (26,27), an interaction that has been postulatedto contribute to the stabilization of the ECM (28,29). Morerecent studies suggest that II components are associated withHA-based cables on colonic smooth muscle cells (30). The patternof expression of TSG-6 and its ligand specificity suggests thatit also may be involved in ECM remodeling (31,32). It is composedof a Link module and a CUB module (residues 37 to 128 and 129to 250, respectively) that are arranged in a contiguous manner,and it has been shown to have a role in regulating cell migration(33–35). TSG-6 binds HA directly and also supports thecovalent transfer of HC of II family members to HA, both ofwhich are likely to stabilize HA matrices. Importantly, Fülöpet al. (27) showed that TSG-6 –/– mice are infertileas a result of their inability to form HA-rich ECM that is essentialfor cumulus expansion, a phenotype that is associated with lackof incorporation of HC of either II or PI into the cumulus matrix.
In this study, we sought to test the hypothesize that an increasein HA synthesis that is driven by the inducible form of HAS,HAS2, may be associated with functional changes in PTC thatcontribute to the pathogenesis of renal injury. For examinationof this hypothesis, this study characterizes HAS2-driven HAsynthesis and its pericellular assembly and how this relatesto alterations in PTC function. The data demonstrate that overexpressionof HAS2 leads to a marked increase in pericellular HA coatsand to HA secretion into the cell culture medium. In contrast,there was a marked decrease in pericellular HA cables. Thesealterations in HA generation were accompanied by increased cellmigration in a scratch wound model and marked reduction in HA-dependentmonocyte binding, which is consistent with a profibrotic PTCphenotype.
Cell Culture and Generation of Cell Lines
All experiments were done using HK-2 cells (CRL-2190; AmericanType Culture Collection, Rockville, MD), which are human PTCthat are immortalized by transduction with human papilloma virus16 E6/E7 genes (36). Cells were cultured in DMEM/Ham’sF12 (Life Technologies, Paisley, UK) supplemented with 10% FBS(Biologic Industries Ltd., Cumbernauld, UK), l-glutamine, insulin,transferrin, sodium selenite, hydrocortisone, and HEPES (pH7.2; Sigma-Aldrich, Poole, UK). Fresh growth medium was addedto cells every 3 to 4 d until confluent. All experiments weredone using cells at passage 30 or below, and cells were growtharrested in serum-free medium for 48 h before use in experiments.All experiments were done in serum-free conditions.
Cell migration was examined as described previously (37,38).Briefly, for quantification of re-epithelialization, an intersectingarea of denuded cells was generated (37,38), and closure ofthe denuded area was monitored using an Axiovert 100M invertedmicroscope fitted with a digital camera (ORCA-1394; HamamatsuPhotonics K.K., Hamamatsu, Japan). Images of the denuded areawere captured as a digitized sequence. The rate of motilityof cells was calculated as the number of cells that enteredthe central denuded area. Cell number was expressed as cellsper mm2 of original denuded area at each specified time point.
U937 cells, originally derived from a human histiocytic lymphoma,were procured from the American Type Culture Collection. Thecells were grown in suspension culture in RPMI medium that wassupplemented with l-glutamine and penicillin/streptomycin andcontained 5% FBS. Cells were routinely subcultured at a 1:5ratio three times per week.
Generation of a HAS2 Overexpressing Clone
HAS2 open reading frame (ORF) was a gift from Dr. Andrew Spicer(Texas A&M University, College Station, TX). Standard PCRwas done to reproduce the ORF using a combination of polymerasesto increase the fidelity of the PCR product as described previously(39). The combination included Taq polymerase (Promega, Madison,WI) and Pfu polymerase (Promega) in a 9:1 ratio. The primersused in the reaction included sites for two different restrictionenzymes (NotI and PstI) to ensure sense cloning of the ORF intothe pcDNA4/TO vector (Invitrogen, Carlsbad, CA). The ORF wasinserted into the vector using a standard ligation reactionwith Promega T4 DNA Ligase. Amplification of the cloned vectorwas done via bacterial transformation (JM109 competent Escherichiacoli; Promega). The integrity of the HAS2 ORF was confirmedby restriction enzyme digestion (NotI and PstI; Figure 1) andsequencing (data not shown).
Figure 1. Confirmation of hyaluronan synthase 2 (HAS2) cloning in sense direction into pcDNA4/TO vector. Purified DNA that was extracted from transformed Escherichia coli and subjected to restriction-enzyme double-digestion (using PstI and NotI) was resolved using 1% (wt/vol) agarose gel electrophoresis. Lanes 1 and 2 show two samples that were cloned successfully with HAS2 open reading frame (1700 bp); lane 3 shows an empty vector that was treated similarly with the restriction enzymes.
To establish a stable transfected cell line, subconfluent HK-2cells in 35-mm Petri dish were transfected with either HAS2-pcDNA4/TOor empty pcDNA4/TO (mock transfection) using FuGENE 6 transfectionreagent (Roche Scientific, Lewes, UK) under serum-free conditions(3 ml of FuGENE 6 and 1 µg of plasmid DNA). Zeocin (400µg/ml) was added to the culture medium (with 10% serum)24 h after transfection. Medium that contained 200 µg/mlZeocin was changed every 72 h until distinct cell clones weremacroscopically visible (usually within 21 d after transfection).Selection of clones was done using cloning discs (Sigma-Aldrich)that were impregnated with trypsin/EDTA. Discs were left onHK-2 clones for no longer than 5 min and then transferred intoa new 35-mm dish (containing 200 µg/ml Zeocin). Cellswere left to grow until they were >50% confluent. Cells thenwere subcultured into 75-cm2 culture flasks from which cellswere either frozen or processed for further experimentation.Transfected cell lines were always kept with 200 µg/mlZeocin except during experiments. Several HAS2-transfected lineswere screened for HA productivity and HAS mRNA levels with oneHAS2-transfected cell line that showed significant differencesin HA levels and HAS2 mRNA compared with mock-transfected cellline considered as a positive HAS2 transfectant and, therefore,used in all of the experiments reported in this work.
Immunocytochemistry
Cells that were cultured in eight-well glass chamber slides(Nunc, Life Technologies/BRL Life Technologies Ltd, Paisley,UK) were analyzed by immunohistochemistry. Cells were grownto confluence and exposed to serum-free conditions for 48 h.Culture medium subsequently was removed, and the cell monolayerwas washed with sterile PBS. Cells were fixed by addition of100% ice-cold methanol for 15 min at –20°C and permeabilizedwith 0.3% (vol/vol) Triton X-100 for 30 min. After fixation,cells were blocked with 50% FBS for 1 h before a further washingstep with PBS. For HA staining, a biotinylated HA-binding protein(b-HABP; Seikagaku Corp., Tokyo, Japan) at 5 µg/ml orthe Q75 (40) rat mAb against human TSG-6 (5.3 µg/ml) thenwas added and incubated at 4°C overnight. The slides werewashed with PBS before incubation with either fluorescence avidin-D(20 µg/ml; Vector Laboratories, Burlingame, CA) or anti-ratTexas red–conjugated antibody (DakoCytomation, Ely, UK)at room temperature for 1 h. After a final washing step, specimenswere affixed to the slides in Vectasheild mounting medium (VectorLaboratories) and analyzed by confocal laser scanning microscopyof the upper surface of the cell monolayer (TCS-40; Leica Microsystems,Cambridge, UK).
Visualization of HA by Exclusion Assay
Cell-associated pericellular matrices were visualized usinga particle exclusion assay (41). In this assay, the medium isremoved from the monolayer cell cultures, and 750 µl ofa suspension of formalin-fixed horse erythrocytes (108/ml 0.1%BSA in PBS) is added to the dishes. Upon settling, the particlesare excluded from zones, or "halos," around the cells, whichare viewed by phase-contrast microscopy.
Determination of HA Concentration
In all experiments, HA concentration in the cell culture supernatantwas determined by an enzyme-linked binding protein assay (HA"Chugai" quantitative test kit; Congenix, Petersborough, UK).Interassay precision ranged from 6.2 to 7.0% (coefficient ofvariation), and intra-assay precision ranged from 3.6 to 4.7%(coefficient of variation). The assay is sensitive to 10 ng/ml,with no cross-reactivity with other glycosaminoglycan compounds.
Flow Cytometry
Cell surface expression of CD44 was assessed by FACS analysis.After detachment of HK-2 cell monolayers, the cells were incubatedwith anti-CD44 common region antibody (Calbiochem, San Diego,CA) for 30 min at 4°C (antibody dilution 1:500). FITC-labeledsecondary antibody (Sigma; dilution 1:100) then was added inFACS buffer (PBS, 10 mM EDTA [Sigma-Aldrich], 15 mM sodium azide[Fisher Chemicals, Loughborough, UK], and 5% BSA [Sigma-Aldrich;pH 7.35]) for 30 min at 4°C. In control experiments, secondaryantibody only was added to the cells. After three washes inFACS buffer, the data were collected using a Becton Dickinson(Oxford, UK) FACSCalibur 4Ca and analyzed using CellQuest Prosoftware.
Western Blot Analysis
For analysis of extracellular signal–regulated kinase(ERK), confluent cells were serum deprived for 48 h and thenlysed by addition of SDS sample buffer (2% [wt/vol]) SDS, 10%[vol/vol] glycerol, 60 mM Tris, and 5% [vol/vol] mercaptoethanol).For analysis of II, confluent cells were serum deprived for48 h before digestion with either Streptomyces hyaluronidase(2 units for 1 h at 37°C) to release matrix molecules thatbound to hyaluronan or subjected to alkali treatment by theaddition of 0.1 M NaOH at room temperature for 10 min. Afterdigestion, the cells were pelleted at 300 x g and 4°C for5 min, and the supernatants were incubated with SDS sample buffer.
Subsequently, Western blot analyses were done by standard methods.Briefly, equal amounts of samples were prepared in SDS samplebuffer and heated at 100°C for 5 min before loading onto10% (wt/vol) SDS-PAGE gels. Samples were electrophoresed inreducing conditions according to the procedure of Laemmli (42).After electrophoresis, the separated proteins were transferredto a nitrocellulose membrane (Amersham Pharmacia, Biotech UKLtd., Buckinghamshire, UK). The membrane was blocked with Tris-bufferedsaline (TBS) that contained 5% (wt/vol) nonfat powdered milkfor 1 h and then incubated with the primary antibodies againstII-related proteins (DakoCytomation; dilution 1:1000), the duallyphosphorylated active form of MAPK (p44/ERK1 and p42/ERK2; Promega;final dilution 1:5000), total MAPK (Sigma; dilution 1:5000)in TBS that contained 1% (wt/vol) BSA, and 0.1% (vol/vol) Tween20 (TBS-Tween) or rat anti-CD44 (Calbiochem; dilution 1:500)overnight at 4°C. The blots subsequently were washed inTBS-Tween and then incubated with an appropriate horseradishperoxidase–conjugated secondary antibody (Sigma) in TBS-Tween.Proteins were visualized using enhanced chemiluminescence (AmershamPharmacia) according to the manufacturer’s instructions.
Assay for Leukocyte Adhesion
U937 cell adhesion was measured as described previously (43).Confluent mock-transfected and HAS2 overexpressing HK-2 cellswere serum deprived for 48 h before the leukocyte adhesion assay.On the day of assay, U937 cells (up to 70 x 106 cells/ml) werelabeled for 90 min at 37°C with 100 µCi 51Cr as sodiumchromate (Amersham BioSciences, Chalford St. Giles, UK). Thelabeled cells were washed three times with serum-free culturemedium, counted on a hemacytometer, and resuspended to 106 viablecells/0.5 ml (as determined by Trypan blue dye exclusion). Incubationmedium was removed from HK-2 cultures, and 106 monocytes wereadded to each well. The binding phase of the assay was doneat 37°C for 1 h. All cultures were washed with cold mediumbefore lysis by 1% Triton X-100. An aliquot subsequently wasremoved for quantification of radioactivity. The number of theU937 cells bound per well was calculated from the initial specificactivity (cpm/cell). Spontaneous release of chromium from U937cells in control incubations on monocytes without HK-2 cellstypically was <10%.
Analysis of 3H-Radiolabeled HA
Confluent monolayers of cells were serum deprived for 48 h beforein vitro3H-labeling of HA with [3H]glucosamine under serum-freeconditions for 72 h as described previously (44). Supernatantsamples were collected and treated with equal volumes of 200µg/ml pronase (Sigma-Aldrich) for 24 h at 37°C forsubsequent analysis of HA that was released into the culturesupernatant. The remaining cell monolayers were incubated with10 µg/ml trypsin (Sigma-Aldrich) in PBS for 10 min atroom temperature to remove pericellular (protein-bound) 3H-HA(Trypsin extract), and an equal volume of 100 µg/ml pronasewas added to the digest for 24 h at 37°C. Finally, aftertrypsinization, 100 µg/ml pronase was added to each cellpellet for 24 h at 37°C to solubilize the remaining cell-associated3H-HA (cell layer).
Each of the fractions subsequently was passed over a DEAE ion-exchangecolumn (Amersham Biosciences) equilibrated with 8 M urea (pH6) in Bis-Tris buffer. The columns were washed using 8 M ureabuffer to remove low molecular weight peptides and unincorporatedradiolabel. HA then was eluted with 0.3 M NaCl in the urea buffer.Equal volumes of eluted HA then were precipitated by 3 volumesof 1.3% (wt/vol) potassium acetate in 95% (vol/vol) ethanolin the presence of 50 µg/ml each of glycosaminoglycan,HA, heparin, and chondroitin sulfate (Sigma-Aldrich) as co-precipitants.Precipitated HA from two equal volumes were either dissolvedin 4 M guanidine HCl buffer or incubated with 1 IU Streptococcalhyaluronidase (ICN Biomedicals, Basingstoke, UK) at 37°Cfor 24 h before addition of equal volumes of 8 M guanidine HClbuffer (pH 6). Each sample was run through a Sephacryl S-500column (Amersham Biosciences) and eluted with 4 M guanidineHCl buffer before quantification of radioactivity in the elutedfractions. The value of the hyaluronidase-treated portion subtractedfrom the non–hyaluronidase-treated portion was taken asa measure of 3H radioactivity in HA.
Alteration in mRNA Expression
The expression of mRNA for HAS, hyaluronidases, and HC3 fromPI was determined by reverse transcription–PCR (RT-PCR)using specific oligonucleotide primers (Table 1) as describedpreviously (45). PCR was done for various cycles (28 to 40 cycles)to ensure that each amplification was in the linear range ofthe curve. After PCR, one tenth of the PCR reaction mixturesfrom both test and control (-actin) samples was mixed and separatedby flatbed electrophoresis in 3% (wt/vol) NuSieve GTG agarosegels (Flowgen Instruments Ltd., Sittingbourne, UK), stainedwith ethidium bromide (Sigma-Aldrich), and photographed. Thephotographic negatives were scanned using a densitometer (Model620 video densitometer; Bio-Rad Laboratories Ltd. Hercules,CA), and the densities of the bands were compared with thoseof the housekeeping gene. Results were expressed as the ratioof the gene of interest to that of -actin, normalized to thecontrol value (the ratio in the unstimulated cells) of eachexperiment.
Expression and Purification of Wild-Type and Mutant TSG-6
Full-length recombinant human TSG-6Q (the protein product ofTSG-6 [46] Q allotype [47]) was expressed in Drosophila S2 cellsand purified as described previously (47). Tyr-94 (numberedas in the preprotein [46]) was mutated to phenylalanine withthe Transformer site-directed mutagenesis kit (Clontech, PaloAlto, CA) using the method described previously (48) and purifiedto homogeneity as for the wild-type protein (M.S.R., C.F., E.Harvey, D.J. Mahoney, C.M. Milner, and A.J.D., manuscript inpreparation). The mutagenesis and selection primers were 5'-CAGAGTTGGATTGCCCATTGTG-3'and 5'-AGAGGGCCCGCGGTTCGAAG-3', respectively, where the basesin bold altered a Tyr to Phe (in the former) and an AvrII restrictionsite to SacII (in the latter). The mutation was confirmed inthe purified protein (TSG-6Q_Y94F) by trypsin digestion andmass spectrometric analysis (M.S.R. et al., manuscript in preparation).The isolated Link module domain from human TSG-6 (Link_TSG-6;corresponding to residues 36 to 133 in the preprotein [46])was expressed in E. coli and purified as before (49).
Effect of TSG-6 Mutation on HC Transfer
Wild-type TSG-6Q and the TSG-6Q_Y94F mutant were compared inan assay that determines the extent of TSG-6–mediatedcovalent transfer of HC from II (HC1 and HC2) and PI (HC3) ontoHA (27). The TSG-6 proteins (250 ng) were incubated with 5 µlof mouse serum (as the source of II and PI) and 5 µg ofhigh molecular mass HA (Healon GV; Advanced Medical Optics Inc.,Santa Ana, CA) in the absence and presence of 2 mM EDTA in 50µl (total volume) of PBS for 24 h at 37°C. The sampleswere run on 4 to 20% (wt/vol) precast gels and analyzed by Westernblotting as described previously (27) using an anti- II polyclonalantibody (Dako Corp, Ely, UK).
Statistical Analyses
Statistical analysis was done using the unpaired t test, witha value of P < 0.05 considered to represent a significantdifference. The data are presented as means ± SD of nexperiments. For each individual experiment, the mean of duplicatedeterminations was calculated.
Overexpression of HAS2 mRNA in the stable cell line was confirmedby RT-PCR (Figure 2A). The specificity of HAS2 mRNA expressionwas demonstrated by examining the expression of HAS3 mRNA inboth the HAS2 overexpressing cell line and a stable cell linethat was transfected with the empty vector (MOCK). By scanningdensitometry, there was approximately a two-fold increase inHAS2 mRNA expression in the stable HAS2 cell line, with no changein HAS3 mRNA as compared with the mock-transfected cell line.
Figure 2. HAS2 overexpression and characterization of HA. (A) HAS2 expression by reverse transcription–PCR. Total mRNA was extracted from confluent monolayers of mock- and HAS2-overexpressing cells after 48 h of serum deprivation. PCR products were separated on a 3% agarose gel and stained with ethidium bromide. Two representative PCR reactions are shown. PCR amplification was performed for 28 cycles for -actin mRNA and 36 cycles for HAS2 mRNA and for HAS3 mRNA. (B) Quantification of HA. In a parallel experiment, supernatant samples were collected from confluent serum-deprived cells and HA was quantified by ELISA. Data represent mean ± SD of four separate experiments. (C) Characterization of HA. Confluent serum-deprived monolayers of HAS2-overexpressing cells () or mock-transfected cells () were exposed to serum-free conditions for 24 h in the presence of 20 µCi/ml [3H]glucosamine. Cell culture supernatant, the trypsin extracts, and cell layer HA fractions were prepared as described in Materials and Methods. Radiolabeled HA subsequently was analyzed by Sephacryl S-500 chromatography.
HA production by the cell lines was examined both by ELISA ofthe cell culture supernatant and also after [3H]glucosaminelabeling of HA. Confluent monolayers of HAS2-expressing cellsor mock-transfected cells were serum deprived for 48 h. Freshserum-free medium subsequently was added for an additional 24h before collection and quantification of HA by ELISA (Figure 2B).The HA concentration in the culture supernatant was significantlygreater in the HAS2-expressing cell line. This represented a70% increase in the HA over the mock-transfected cells.
Analysis on Sephacryl S-500 of the [3H]glucosamine-labeled HAsamples from both the HAS 2 overexpressing cell line and themock-transfected HK-2 cells demonstrated that the majority ofthe labeled HA in the supernatant and the trypsin extracts appearednear the void volume and therefore was considered to be of highmolecular mass (Figure 2C). In the HAS2-transfected cells, therewas a marked increase in high molecular weight HA in the mediumand cell layer but not in the trypsin extract. There also wasa marked increase in lower molecular weight HA in the cell layerextract as compared with the mock-transfected cells.
Visualization of Pericellular HA
Confocal imaging was used to examine the organization of HAon the cell surface. HA was identified with b-HABP and detectedwith fluorescence avidin-D. Photomicrographs of fixed growth-arrestedmock-transfected HK-2 cells revealed diffusely arranged pericellularHA over the cell surface (pericellular coats). In addition,HA was demonstrated in cable-like structures that spanned severalcell lengths (Figure 3A). These results are consistent withour previous characterization of cell surface HA in unmanipulatedHK-2 cells (21). In the HAS2 overexpressing cell line, therewas an increase in the size of the pericellular HA coats asassessed by immunohistochemistry (Figure 3B) and an increasein the pericellular HA halo assessed by exclusion of formalin-fixedhorse erythrocytes (Figure 3, C and D). In contrast to the increasepericellular coat, the overexpression of HAS2 resulted in inhibitionof HA cable formation. Confirmation of the nature of the HAcontent of the cable structures was shown by treatment of confluentmonolayers of cells with bovine testicular hyaluronidase (200µg/ml final concentration at room temperature for 5 min)before addition of b-HABP. Limited hyaluronidase digestion removedboth the cable structures in the mock-transfected cells andthe pericellular matrices in both the mock- and HAS2-transfectedcells (Figure 3, A and B inserts).
Figure 3. Visualization of HA on the surface of a cell monolayer. Confluent monolayers of mock (A) and HAS2-overexpressing (B) cells were serum deprived for 48 h before fixation with methanol and detection of HA by addition of biotinylated HA-binding protein (b-HABP). Monolayers were imaged by confocal microscopy (x10 objective). Pericellular coats are indicated by red arrowheads, and HA cables are indicated by white arrows. For confirmation of the nature of HA staining, in parallel experiments, cells were treated with bovine testicular hyaluronidase (final concentration 200 µg/ml) at 37°C for 5 min, before fixation and addition of b-HABP (+H’ase inserts). Functional pericellular matrix in the mock (C) and HAS2-overexpressing (D) cells was visualized by the particle exclusion assay. The zone of exclusion of formalin-fixed red blood cells in both cell lines is indicated by arrowheads.
Functional Alterations Associated with HAS2 Overexpression
Two functional parameters that we previously demonstrated tobe related to HA expression are the binding of monocytes andepithelial cell migration. Monocyte binding depends on HA cellsurface cables (21), whereas increased HA synthesis in a scratch-woundhealing model is associated with increased cell migration (50).These two functional parameters were chosen because bindingof monocytes attenuates monocyte-stimulated TGF-1 generationand may be considered as a disease-limiting function, whereasmigration of PTC is associated with transdifferentiation anda profibrotic function.
HA-Dependent Monocyte Binding
U937 monocytic cells were used to examine the binding capacityof inflammatory cells by either HAS2-overexpressing or mock-transfectedHK-2 cells. Quantification of monocyte binding was done by determinationof bound radioactivity after addition of 51Cr-labeled U937 cells.Despite increased HA production in the HAS2-overexpressing cells,binding of labeled U937 cells was significantly greater in themock-transfected cells (Figure 4). In this regard, monocytebinding was significantly reduced by removal of cell surfaceHA by adding bovine testicular hyaluronidase (final concentration200 µg/ml) at 37°C for 5 min to the epithelial cellmonolayers, before addition of monocytes (Figure 4A). Previously,we demonstrated that residual monocyte binding after hyaluronidasetreatment (and removal of HA cables) represents binding to intracellularadhesion molecule expressed on the PTC cell surface, whereasHA-dependent binding represents binding to HA cables (21,51).Addition of hyaluronidase to the HAS2 cell line did not significantlyreduce the binding of labeled U937 cells. These data thereforeare consistent with the marked difference in the expressionof HA cables between the two cell lines.
Figure 4. Reduced HA cable generation is associated with reduced monocyte binding and increased expression of hyaluronidases in the HAS2-overexpressing cells. (A) Quantification of monocyte binding. Confluent monolayers of serum-deprived cells were washed with PBS before addition of 1 x 106 51Ci chromium-labeled U937 cells again under serum-free conditions for 1 h at 37°C. Quantification of bound radioactivity was done as described in Materials and Methods. For quantification of HA-dependent binding, the monolayer was treated with bovine testicular hyaluronidase (H’ase; final concentration 200 µg/ml) at 37°C for 5 min before addition of monocytes. Data represent mean ± SD of four individual experiments. (B) Endogenous expression of human hyaluronidase (hyal1 and hyal2) mRNA. Total mRNA was extracted from confluent monolayers of mock and HAS2-overexpressing cells after 48 h of serum deprivation. PCR products were separated on a 3% (wt/vol) agarose gel and stained with ethidium bromide. Results from RNA extraction from three separate cell cultures are shown. PCR amplification was done for 28 cycles for -actin mRNA, 36 cycles for hyal1 mRNA, and 32 cycles for hyal2 mRNA. Densitometric ratios of the gene of interest (either hyal 1 [lanes 1 and 2] or hyal 2 [lanes 3 and 4]) compared with the housekeeping gene -actin of three individual experiments are shown with the data representing mean ± SD.
Previously, we demonstrated that the increase in HA cables isassociated with a downregulation of hyaluronidase, suggestingthat alteration in HA turnover may be involved in their generation.No commercial antibodies are available to examine hyaluronidaseprotein expression; therefore, to explore this possibility further,we examined hyaluronidase expression (hyal1 and hyal2) by descriptiveRT-PCR analysis in the HAS2- and mock-transfected cell lines.Confluent cell monolayers were serum deprived for 48 h beforeextraction of total mRNA. The marked decrease in HA cables inthe HAS2-overexpressing cell line and the increase in low molecularweight HA in the cell extract were associated with a significantincrease in both hyal1 and hyal2 mRNA expression (Figure 4B).
Cell Migration
Cell migration was assessed in a previously characterized scratch-woundsystem (13,50). Confluent monolayers of cells were serum deprivedfor 48 h before generation of an intersecting area of denudedcells by scraping with a sterile 1000-µl pipette tip.Closure of the denuded area then was monitored at differenttimes. At 24 h and all time points beyond this, the number ofmigrating cells that entered the "denuded area" was significantlygreater in the HAS2-overexpressing cells compared with the mock-transfectedcells (Figure 5).
Figure 5. Quantification of cell migration. Confluent serum-deprived monolayers of mock () or HAS2-overexpressing cells () were scratched as described in Materials and Methods to produce an intersecting area that was denuded of cells. Subsequently, after washing of the monolayer to remove detached cells, the rate of cell migration of each of the two cell lines was assessed by directly counting the number of cells that migrated into the intersecting denuded area at each of the time points indicated. The data are expressed as the number of cells per mm2 of denuded area. Data represent the mean ± SD of four individual experiments.
Mechanism of Enhanced Migration and Involvement of Hyaladherins: CD44, II, PI, and TSG-6 CD44.
Previously, we demonstrated that migration in a scratch-woundsystem that used nontransfected HK-2 cells is mediated by CD44-dependentactivation of ERK and that this effect may be enhanced by theaddition of exogenous HA (50). In contrast, in our study, additionof monoclonal blocking antibody to CD44 (Clone BU52; The BindingSite Ltd., Birmingham, UK; final concentration 5 µg/ml)that blocks HA–CD44 interactions (50), to a monolayerof HAS2-transfected cells after the creation of an intersecting"scratch" had no effect on the rate of cell migration as comparedwith migration of HAS2-overexpressing cells in the absence ofantibody (Figure 6A). Cell surface CD44 expression was examinedby flow cytometry using an antibody to the common region ofCD44 (Calbiochem). This revealed that there was a marked decreasein cell surface expression of CD44 in the HAS2-overexpressingcells compared with the mock-transfected cells (Figure 6B),which also was confirmed by Western blot analysis of total celllysates (Figure 6C). Furthermore, reduction in the ERK activationwas observed as assessed by immunoblot analysis of phosphorylatedMAPK in the HAS2-overexpressing cells as compared with mock-transfectedcells (Figure 6D).
Figure 6. Enhanced migration is CD44 independent and associated with reduced CD44 expression in HAS2-overexpressing cells. (A) Blocking of CD44 with an anti-CD44 mAb did not reduce migration of HAS2-overexpressing cells. Confluent serum-deprived monolayers of HAS2-overexpressing cells were scratched as described in Materials and Methods to produce an intersecting area that was denuded of cells. Subsequently, after washing of the monolayer to remove detached cells, the rate of cell migration was assessed in the presence () or absence () of blocking antibody to CD44 (5 µg/ml) as described in legend to Figure 5. The data are expressed as the number of cells per mm2 of denuded area at each time point as indicated. Data represent the mean ± SD of four individual experiments. (B) Confluent monolayers of mock () or HAS2-overexpressing cells () were serum deprived for 48 h before detachment. Cells were incubated with anti-CD44 common region antibody, and cell surface expression of CD44 was assessed by FACS analysis. (C) In parallel experiments, CD44 protein expression was examined by Western blot analysis of cell lysates that were extracted from serum-deprived HK2 cells (lane 1), mock-transfected cells (lane 2), or HAS2-overexpressing cells (lane 3). (D) Activation of extracellular signal–regulated kinase (ERK) was used as a surrogate marker of CD44 activity associated with cell migration. Whole-cell extracts from confluent growth-arrested mock or HAS2-expressing cells were subjected to SDS-PAGE and Western blotting. The blots were probed with antiactive mitogen-activated protein kinase (MAPK) antibody, which recognizes the dually phosphorylated active form of MAPK. After stripping of the membrane (20% SDS, 50°C for 30 min), the blot was reprobed with an anti–total MAPK antibody to ensure no change in the expression of whole-cell MAPK had occurred. (E) HAS2 cells exhibit enhanced migration when stimulated with HA. After generation of an intersecting area that was denuded of cells, fresh cell culture supernatant alone () or supernatant to which HA (molecular weight 2 x 106) had been added to a final concentration of 25 µg/ml () was added and the rate of cell migration was assessed by direct counting of the number of cells that migrated into the intersecting denuded area at each of the time points indicated. The data are expressed as the number of cells per mm2 of denuded area and represent the mean ± SD of four individual experiments.
To demonstrate that migration in the HAS2 cells was relatedto the generation of HA, we added exogenous HA (molecular weight2 x 106, 25 µg/ml, generated as described previously [50])to the monolayer after the generation of a wound/denuded area.Despite reduction in the expression of CD44, the migratory responseof the HAS2-overexpressing cells was significantly enhancedby addition of exogenous HA at all time points beyond 48 h (Figure 6E).
II/PI.
Although HA is known to be the major structural component ofthe pericellular matrix, other macromolecules, including IIand PI, are important in its organization. After scratch wounding,incubation of either mock- or HAS2-transfected cells with antibodyto II/PI led to significant inhibition of cell migration (Figure 7).Previously, we demonstrated that PTC synthesize HC3 of the PIcomplex but not HC1 of II (24). Expression of HC3 mRNA and proteinwas examined in the HAS2-overexpressing cells by RT-PCR andWestern blot analysis. In the HAS2-overexpressing cells, therewas a decrease in HC3 mRNA expression compared with mock-transfectedcells (Figure 8A). To investigate further the presence of II-relatedproteins in the cell matrix, we isolated total cell extractsby Laemmli SDS buffer under reducing conditions (total extract)and by treatment with Streptomyces hyaluronidase. The extractswere analyzed by SDS-PAGE under reducing conditions followedby immunoblotting and immunochemical detection with a polyclonalantiserum against II (that also recognizes intact PI [52]).After hyaluronidase digestion, two major II-positive bands wherethe lower species (approximately 75 kD) most likely representsHC that were covalently bound to HA were observed (Figure 8B).The appearance of the upper band (approximately 125 kD) on hyaluronidasedigestion and its sensitivity to NaOH treatment are similarto that described for the TSG-6–HC complexes in the murinecumulus-oophorus (53).
Figure 7. Inhibition of cell migration by antibody to inter--inhibitor (II)/pre–-inhibitor (PI) in both mock and HAS2-overexpressing cells. After generation of an intersecting area that was denuded of cells, the rate of cell migration of each of the two cell lines was assessed at each of the time points indicated in the presence and absence of antibody to II/PI as indicated (dilution 1:50). The data are expressed as the number of cells per mm2 of denuded area. Data represent the mean ± SD of four individual experiments.
Figure 8. Expression of heavy chain 3 (HC3) chain and incorporation of HC into the matrix of HAS2-overexpressing cells. Total mRNA was extracted from confluent monolayers of mock and HAS2-overexpressing cells after 48 h of serum deprivation (A). PCR products were separated on a 3% (wt/vol) agarose gel and stained with ethidium bromide. Three representative PCR reactions are shown. PCR amplification was done for 28 cycles for -actin mRNA and 36 cycles for PI HC3 chain mRNA. Densitometric ratios of the gene of interest compared with the housekeeping gene, -actin, of three individual experiments are shown with the data representing mean ± SD. In parallel experiments, expression of II/PI was examined by Western blot analysis (B). Total cell extracts (Total) were prepared by addition of SDS sample buffer. HA-bound proteins were release from the total cell extract by Streptomyces hyaluronidase digestion (H’ase). In addition, total cell extracts were subjected to alkali treatment by the addition of 0.1 M NaOH. The total cell extracts or the supernatant samples after enzymatic digestion or alkali treatment were analyzed by SDS-PAGE, and II-related proteins were detected by Western blot analysis using a polyclonal antibody to II.
TSG-6.
TSG-6 is a secreted HA-binding protein that is implicated inECM remodeling (33). It interacts with HA and can form withHC of II and PI covalent complexes that act as intermediatesin the transfer of HC from II/PI to HA (54). The presence ofendogenous TSG-6 in the pericellular matrix was confirmed byimmunohistochemistry, which revealed that TSG-6 co-localizedwith HA around the HAS2-overexpressing cells in monolayer culture(Figure 9). TSG-6 also localized to the smaller pericellularmatrix coats in mock-transfected cells. In our scratch-woundsystem, incubation of HAS2-overexpressing cells with the ratanti-human mAb A38 (final concentration 5 µg/ml), previouslydemonstrated to block HA binding to TSG-6 (40) and inhibit theformation of TSG-6–HC complexes in vitro (55), significantlyattenuated cell migration (Figure 10), whereas it did not affectmock-transfected cells.
Figure 9. TNF-–stimulated gene 6 (TSG-6) co-localizes to the HA pericellular matrix of HAS2-overexpressing cells. Confluent monolayers of HAS2-overexpressing cells (A, C, and D) or mock-transfected cells (B) were serum deprived for 48 h before fixation with methanol and detection of either HA by addition of b-HABP (5 µg/ml; C) or rat anti-human TSG-6 mAb Q75 (5.3 µg/ml; A and B). After overnight incubation, the slides were washed with PBS before incubation with either fluorescence avidin-D (20 µg/ml) or anti-rat Texas red–conjugated antibody. Co-localization of HA and TSG-6 was examined by merging of the two images (D).
Figure 10. Inhibition of TSG-6 inhibits migration of HAS2-overexpressing cells but not mock-transfected cells. After generation of an intersecting area that was denuded of cells, the rate of cell migration of each of the two cell lines was assessed at each of the time points indicated in the presence and absence of rat anti-human mAb A38 as indicated (final concentration 5 µg/ml). The data are expressed as the number of cells per mm2 of denuded area. Data represent the mean ± SD of four individual experiments.
To further investigate the potential mechanism by which TSG-6influenced migration in the HAS2-overexpressing cells, we assessedmigration in the presence of fully functional full-length recombinantTSG-6Q, Link_TSG-6, which can bind to HA (35,47,56) but is unableto form stable complexes with HC (49), and a mutant of TSG-6Q,in which Tyr-94 has been replaced by phenylalanine (TSG-6Q_Y94F),resulting in impaired HA binding (48). In this study, we assessedthe effect of this mutation on TSG-6–mediated HC transferonto HA (i.e., to form HC–HA). Figure 11A shows that TSG-6Qand TSG-6Q_Y94F both can support HC transfer from II and PIonto HA as shown by the loss of II and PI in lanes 1 and 3.Both TSG-6–HA formation and subsequent HC transfer areknown to be metal ion dependent (54); consistent with this,the transfer activity of the mutant was inhibited by EDTA. Inthe scratch-wound model, increased migration of HAS2-overexpressingcells was seen after addition of either TSG-6Q or TSG-6Q_Y94F(Figure 11B). In contrast, addition of Link_TSG-6, which canbind to HA (35,47,56) but is unable to form stable complexeswith HC (49), had no effect on migration. These data suggestthat TSG-6–mediated HC transfer is important for enhancedmigration of cells in which HA production is increased throughoverexpression of HAS2 mRNA.
Figure 11. (A) Effect of mutation on TSG-6–mediated transfer of HC onto HA. Mouse serum (as a source of II and PI), high molecular mass HA, and human recombinant TSG-6Q (wild-type [wt] or TSG-6Q_Y94F mutant [Y94F]) were incubated at 37°C in the absence or presence of EDTA. Samples were run on SDS-polyacrylamide gels and analyzed by Western blotting using a polyclonal antibody against II that also recognizes PI. The presence of EDTA (+) inhibits HC transfer, and intense bands are seen for II and PI, as described previously (54); the absence of a band or a significant reduction in its intensity indicates that TSG-6–mediated HC transfer has taken place. TSG-6Q and Y94F in the absence of EDTA (–) both give rise to a high level of HC transfer. However, if TSG-6 is not added, then transfer does not occur. Whereas the Y94F mutant leads to a significant amount of HC transfer, it is somewhat less active than the wt protein for the transfer of II and PI HC onto HA. (B) Enhanced migration of HAS2-overexpressing cells depends on II/PI transfer activity of TSG-6. After generation of intersecting area that was denuded of cells, the rate of cell migration was assessed at each of the time points indicated in the presence of recombinant full-length human TSG-6 (Q allotype; TSG-6Q), a mutant of the full-length protein TSG-6Q_Y94F, the isolated TSG-6 Link module domain (Link_TGS6; all at a final concentration of 5 µg/ml). The data are expressed as the number of cells per mm2 of denuded area. Data represent the mean ± SD of four individual experiments.
HA is a large glycosaminoglycan that is present in most ECM(57). It is not a prominent component of the normal renal cortex,but increased levels have been demonstrated after renal injury,although the functional significance of this observation isnot clear. We previously demonstrated that exogenous HA mayactivate PTC and enhance their migration in a scratch model(16,50). Recent work has demonstrated that increased HA productionby renal epithelial cells induces mesenchymal transdifferentiation(58), a key event that contributes to the accumulation of renalinterstitial fibroblasts that are associated with progressiverenal fibrosis. We also demonstrated that PTC assemble HA-basedcables, which bind monocytes and reduce their capacity to interactwith cell surface adhesion molecules (21). It is clear, therefore,that alterations in HA synthesis and organization will haveimportant roles in events that contribute to the pathogenesisof progressive renal fibrosis.
Unlike other glycosaminoglycans, HA normally is synthesizedat the plasma membrane rather than in the Golgi and is thoughtto be elongated at the reducing rather than the nonreducingterminus, obviating the need for a core protein (59). Threeseparate HAS genes have been cloned and characterized in humans:HAS1, HAS2, and HAS3 (17,60). The three HAS genes are locatedon separate autosomes, are expressed in different patterns duringdevelopment (and in the adult), and are thought to be subjectto different regulatory influences. Our previous studies demonstratedthat in PTC, stimulation of HA is associated with inductionof the HAS2 gene. Furthermore, phenotypic alteration in renalepithelial cells may be driven by adenoviral expression of HAS2(58). The aim of the work presented in this study was to characterizealterations in HA and its consequences on cell functions afterHAS2-dependent HA synthesis in PTC to examine the potentialrole of HAS2-driven HA synthesis in the alteration of epithelialcell function.
The data presented demonstrate a striking absence of HA cablesafter overexpression of HAS2 in PTC when compared with the mock-transfectedcells. Previously, we demonstrated that HA cables are associatedwith monocyte adhesion, which is consistent with the data presentedin this study that the HAS2-overexpressing cells exhibit a reductionin HA-dependent monocyte binding. HA cable-like structures bindmononuclear leukocytes via their cell surface CD44 receptors(21). We demonstrated that binding of monocytes to these structuresattenuates monocyte-dependent PTC generation of TGF-1 (22) andtherefore may be disease limiting because TGF-1 is well establishedas a mediator of progressive renal fibrosis. That HA cable formationthat is inhibited by HAS2-overexpression therefore is consistentwith our previous observations that stimuli that generally areimplicated in disease pathogenesis also stimulate HAS2 geneexpression (18) and suggests that this inducible HAS isoformtherefore may contribute to renal injury.
Previously, we demonstrated that the stimulation of cable formationafter addition of bone morphogenic protein 7 (BMP-7) was associatedwith decreased expression of hyal1 and hyal2 (21). In contrastto BMP-7, IL-1 did not influence hyal expression and did notstimulate cable formation, although it is a potent stimulusof HA synthesis. This led us to postulate that downregulationof hyaluronidase activity by BMP-7 allowed HA to remain associatedwith HAS and that HA that is extruded through the cell membraneis anchored to HAS isoforms and associates with similarly anchoredHA from neighboring cell, thereby forming cables. Our observationsin this study are consistent with this hypothesis; a consequenceof the upregulation of hyaluronidase in the HAS2-transfectedcells is that extruded HA may be cleaved at the cell surface.Although we have demonstrated only alteration of hyal1 and hyal2at the level of their mRNA, the increase in low molecular weightHA in the HAS2 cells suggests that this also is associated withincreased enzyme activities. Increased hyaluronidase expression,however, may represent a feedback response of the cell whenit encounters a huge excess of HA and therefore may not be causallyrelated to absence of cables in HAS2-overexpressing cells. Analternative explanation for absence of cables may be relatedto organization of HAS2 and that competition for the UDP-sugarsubstrates between cable-forming and non–cable-formingHAS2 may limit cable formation. It is possible to speculatethat cable-forming HAS may have specific membrane organizationthat requires additional structural components that are notpresent in the transfected cells, which therefore limits cableformation in these cells. A third possibility is that cablesmay be associated with the activity of the constitutive HASisoform HAS3. Transfection with HAS2 therefore may lead to competitionfor substrate, attenuating HAS3-associated HA generation andlimiting cable formation.
The other striking feature of the HAS2-overexpressing cellsis the increase in the pericellular HA coat. This is consistentwith the observation that inhibition of HAS2 activity has beendemonstrated to inhibit pericellular matrix assembly in humanarticular chondrocytes (61). The generation of such a pericellularmatrix has been associated with the migratory capacity of cells(19), as was seen in the HAS2-overexpressing cells in our study.Previously, we demonstrated that exogenous HA increased epithelialcell migration through CD44-dependent activation of the ERK-MAPKpathway. In this study, enhanced migration was not dependenton CD44 activation. Furthermore, we were able to demonstratea reduction of cell surface expression of CD44 in the HAS2-overexpressingcells and a reduction of expression of CD44 in total cell lysates.This may suggest that in contrast to previous studies that usedarticular chondrocytes (41), the pericellular HA matrix in theseepithelial cells may not be anchored by CD44 in this case. Incertain cell types, it has been shown that HA is internalizedfor degradation by an endocytic pathway that requires CD44 function(62,63). It is possible that reduced cell surface CD44 expressionin the HAS2-overexpressing cells may be due to increased HAinternalization. This is supported by the observation of increasedamounts of intracellular HA of low molecular weight in the transfectedcells. Our observations also are consistent with reports thatHA-dependent cell migration may be CD44 independent but associatedwith another distinct HA cell surface receptor, namely receptorfor HA-mediated motility (RHAMM) (64,65). More recent studiessuggested that RHAMM compensates for CD44 in CD44 knockout mice(66). It is interesting to speculate, therefore, that in theface of reduced CD44 expression in the HAS2-overexpressing epithelialcells, enhanced migration is related to preferential RHAMM activation.It is of note, however, that HA-induced transcription of metalloproteaseshas been demonstrated to be mediated via a cell surface receptorthat is neither CD44 nor RHAMM (67), suggesting that as-yet-unidentifiedcell surface receptors may mediate HA-dependent events.
In contrast to the lack of effect of the blocking antibody toCD44, antibodies to either II-related proteins or TSG-6 clearlyattenuated migration of HAS2-overexpressing cells. The localizationof HC within the pericellular HA coat and the inhibition ofmigration of both mock- and HAS2-transfected cells by inclusionof an antibody to II/PI support the notion that addition ofHC to HA is an important mechanism by which the pericellularHA coat is stabilized in renal epithelial cells. Previous studieshave demonstrated the importance of HC in formation of bothHA cables (30) and pericellular HA matrices (27). The lack ofHA cables in the HAS2-expressing cells therefore suggests thatthe effect of blocking II is likely to be related to inhibitionof HC transfer into the pericellular coat. This is supportedfurther by the data demonstrating the involvement of TSG-6 incell migration, as it has been shown that TSG-6 is not requiredfor HA cable formation (68). TSG-6 and its Link module domainwere shown previously to influence leukocyte migration by enhancingtheir adhesion through increasing the affinity of HA bindingto CD44 on the surface of lymphocyte cell lines (35); this mayoccur by the TSG-6–mediated formation of cross-linkedHA fibers that can lead to the clustering of multiple CD44 molecules.The lack of effect of addition of Link_TSG-6 in our system suggeststhat its action is independent of its HA-binding capacity, and,again, this is consistent with the decreased expression of CD44.
TSG-6 also forms covalent complexes with HC from either II (HC1or HC2) or PI (HC3) that act as intermediates in their covalenttransfer onto HA (54), thereby stabilizing ECM that are richin HA (27). There seem to be differences in the mechanism oftransfer of HC1/HC2 and HC3, however, because TSG-6–HCcomplexes that are derived through HC transfer from II can begenerated in vitro in the absence of serum (54), whereas TSG-6–HC3complexes are formed only in vitro in the presence of serum(27,54). It has been proposed that the Link module of TSG-6is involved directly in the formation of TSG-6–HC complexesbecause the A38 antibody, which has its epitope in the Linkmodule (25), can inhibit this process and, consequently, blockHC transfer onto HA (54,55). However, the Link module aloneis unable to form covalent complexes with HC (49). In contrastto Link_TSG-6, a mutated variant of the full-length protein,which has impaired HA binding but still can mediate the transferof HC from II and PI, was able to augment cell migration inour system. These data therefore support the hypothesis thatTSG-6–mediated transfer of HC into the HA-rich pericellularmatrix is a critical regulator of cell migration.
Although the components of the II family are synthesized predominantlyin hepatocytes, we previously demonstrated that PTC cells expressPI, a member of the II family (24). In our cell culture model,cells were grown to confluence in the presence of serum. Duringthis period of growth, HC may be incorporated into the matrixfrom serum-derived II or PI or, alternatively, from endogenousPI. We speculate, however, that the HC that are incorporatedduring the period of migration after scratch wounding are derivedfrom endogenous PI because all of the experimental manipulationswere performed under serum-free conditions. It is interestingthat in our experimental system, the antimigratory effect ofblocking TSG-6 was apparent only in the HAS2-overexpressingcells. It is of note, therefore, that there was decreased expressionof HC3 chains in the HAS2-overexpressing cells. In these cells,given the relative deficiency of HC, the presence of TSG-6,which co-localizes to the HA in the pericellular coat, may bemore critical for the formation of a stabilized pericellularmatrix.
We postulate that the form in which pericellular HA is assembledand, hence, its effect on cell function are determined by acomplex series of mechanisms. The data are consistent with thehypothesis that HA synthesis that results from activity of theinducible HAS isoform HAS2 may generate a profibrotic phenotype.However, in addition to HAS activity, it is likely that theexpression patterns of hyaluronidases and also the generationand expression patterns of various hyaladherins play a pivotalrole in the assembly of pericellular HA.
Acknowledgments
This work was supported in part by a research grant from theNational Kidney Research Fund. A.O.P. is supported by a GlaxoSmithKlineAdvanced Fellowship. The Institute of Nephrology is supportedby Kidney Wales Foundation. A.J.D. acknowledges the supportof the arthritis research campaign (grants 16119 and 16539)and the Medical Research Council.
Footnotes
Published online ahead of print. Publication date availableat www.jasn.org.
Hansell P, Goransson V, Odlind C, Gerdin B, Hallgren R: Hyaluronan content in the kidney in different states of body hydration. Kidney Int 58
: 2061–2068, 2000[CrossRef][Medline]
Sibalic V, Fan X, Loffing J, Wurthrich RP: Upregulated renal tubular CD44, hyaluronan, and osteopontin in kdkd mice with interstitial nephritis. Nephrol Dial Transplant 12
: 1344–1353, 1997[Abstract/Free Full Text]
Lewington AJP, Padanilam BJ, Martin DR, Hammeramn MR: Expression of CD44 in kidney after acute ischemic injury in rats. Am J Physiol 278
: R247–R254, 2000
Wells A, Larsson E, Hanas E, Laurent T, Hallgren R, Tufvesson G: Increased hyaluronan in acutely rejecting human kidney grafts. Transplantation 55
: 1346–1349, 1993[Medline]
Wells AF, Larsson E, Tengblad A, Fellstrom B, Tufvesson G, Klareskog L, Laurent TC: The localisation of hyaluronan in normal and rejected human kidneys. Transplantation 50
: 240–243, 1990[Medline]
Sano N, Kitazawa K, Sugisaki T: Localization and roles of CD44, hyaluronic acid and osteopontin in IgA nephropathy. Nephron 89
: 416–421, 2001[CrossRef][Medline]
Yang J, Liu Y: Dissection of key events in tubular epithelial to myofibroblast transition and its implications in renal interstitial fibrosis. Am J Pathol 159
: 1465–1475, 2001[Abstract/Free Full Text]
Tian YC, Phillips AO: Interaction between the TGF-beta type II receptor/Smad pathway and beta-catenin during TGF-beta1 mediated adherens junction disassembly. Am J Pathol 160
: 1619–1628, 2002[Abstract/Free Full Text]
Evans RA, Tian YC, Steadman R, Phillips AO: TGF-beta1-mediated fibroblast-myofibroblast terminal differentiation-the role of Smad proteins. Exp Cell Res 282
: 90–100, 2003[CrossRef][Medline]
Grande M, Franzen A, Karlsson JO, Ericson LE, Heldin NE, Nilsson M: Transforming growth factor-beta and epidermal growth factor synergistically stimulate epithelial to mesenchymal transition (EMT) through a MEK-dependent mechanism in primary cultured pig thyrocytes. J Cell Sci 115
: 4227–4236, 2002[Abstract/Free Full Text]
Zeisberg M, Bonner G, Maeshima Y, Colorado P, Muller GA, Strutz F, Kalluri R: Collagen composition and assembly regulates epithelial mesenchymal transdifferentiation. Am J Pathol 159
: 1313–1321, 2001[Abstract/Free Full Text]
Tian YC, Phillips AO: TGF-beta1 mediated inhibition of HK-2 cell migration. J Am Soc Nephrol 14
: 631–640, 2003[Abstract/Free Full Text]
Ito T, Williams JD, Fraser DJ, Phillips AO: Hyaluronan regulates TGF-beta1 receptor compartmentalisation. J BiolChem 279
: 25326–25332, 2004
Di Guglielmo GM, Le Roy C, Goodfellow AF, Wrana JL: Distinct endocytic pathways regulate TGF-beta receptor signalling and turnover. Nat Cell Biol 5
: 410–421, 2003[CrossRef][Medline]
Ito T, Williams JD, Fraser DJ, Phillips AO: Hyaluronan attenuates TGF-beta1 mediated signalling in renal proximal tubular epithelial cells. Am J Pathol 164
: 1979–1988, 2004[Abstract/Free Full Text]
Spicer AP, McDonald JA: Characterisation and molecular evolution of a vertebrate hyaluronan synthase gene family. J BiolChem 272
: 1923–1932, 1998
Jones SG, Jones S, Phillips AO: Regulation of renal proximal tubular epithelial cell hyaluronan generation: Implications for diabetic nephropathy. Kidney Int 59
: 1739–1749, 2001[CrossRef][Medline]
Evanko SP, Angello JC, Wight TN: Formation of hyaluronan- and versican-rich pericellular matrix is required for proliferation and migration of vascular smooth muscle cells. Arterioscler Thromb Vasc Biol 19
: 1004–1013, 1999[Abstract/Free Full Text]
Cohen M, Klein E, Geiger B, Addadi L: Organization and adhesive properties of the hyaluronan pericellular coat of chondrocytes and epithelial cells. Biophys J 85
: 1996–2005, 2003[Abstract/Free Full Text]
Selbi WD, De La Motte CA, Hascall VC, Phillips AO: BMP-7 modulates HA mediated proximal tubular cell-monocyte interaction. J Am Soc Nephrol 15
: 1199–1211, 2004[Abstract/Free Full Text]
Zhang XL, Selbi WD, De La Motte CA, Hascall VC, Phillips AO: Renal proximal tubular epithelial cells transforming growth factor beta1 generation and monocyte binding. Am J Pathol 165
: 763–773, 2004[Abstract/Free Full Text]
Salier JP, Rouet P, Raguenez G, Daveau M: The inter-alpha inhibitor: From structure to regulation. Biochem J 315
: 1–9, 1996
Janssen U, Thomas G, Glant T, Phillips AO: Regulation of inter-alpha-trypsin inhibitor (IalphaI) and tumour necrosis factor-stimulated gene 6 (TSG-6) expression in human renal proximal tubular epithelial cells. Kidney Int 60
: 126–136, 2001[CrossRef]
Top
Abstract
Introduction
-inhibitor (I
1 in progressive renal fibrosis. It is particularly important in regulation of cell–cell adhesion (8) and regulation of epithelial and fibroblast cell phenotype (9,10). Synergistic effects with other cytokines (11) and also alterations extracellular matrix (ECM) (12) are known to facilitate these changes. We previously demonstrated that TGF-
) or mock-transfected cells (
) were exposed to serum-free conditions for 24 h in the presence of 20 µCi/ml [3H]glucosamine. Cell culture supernatant, the trypsin extracts, and cell layer HA fractions were prepared as described in Materials and Methods. Radiolabeled HA subsequently was analyzed by Sephacryl S-500 chromatography.
) or HAS2-overexpressing cells (
) were scratched as described in Materials and Methods to produce an intersecting area that was denuded of cells. Subsequently, after washing of the monolayer to remove detached cells, the rate of cell migration of each of the two cell lines was assessed by directly counting the number of cells that migrated into the intersecting denuded area at each of the time points indicated. The data are expressed as the number of cells per mm2 of denuded area. Data represent the mean ± SD of four individual experiments.
